D. Applications of Simulation for Integrated Logistics
1. Simulation and Manufacturing [SS]
In recent years interest in simulation modeling has greatly increased. Many manufacturing companies are investing in advanced manufacturing technology including FMS, various CAM systems, etc. The availability of a well constructed and validated computer model allows the systems designers, the engineers and the managers to understand in advance the detailed consequences of their decisions and the investments prior to actually making binding commitments [9].
The following article discusses the current practices carried out by the manufacturing industries in the field of simulation. It presents a broad picture of different scenario by means of examples and how major obstacles have been overcome by means of simulation [10].
THE PRACTICAL USES OF 3D SIMULATION IN INDUSTRY
The use of 3D graphical simulation techniques can result in substantial savings in work cell design and implementation within the manufacturing industry. This article describes how the GRASP simulation package is a powerful tool in assisting decision making for management. Its application areas include feasibility study, visualization, and evaluation of different equipment and optimization of work layout [5].
Advantages of 3D Simulation [5]
- 2D drawings available to the engineers are difficult to produce and are time consuming to analyze. The results are often inaccurate and ambiguous. 3D simulation frees the engineer from these constraints allowing him to model the complete system designs and simulate the final system operation using computer techniques.
- A full 3D-modeler and a Kinematic modeler, allow complete simulation of a wide range of industrial robots, special purpose manipulators, and other jointed structures. [Ref4]
- High level programming language and sophisticated logical constructs allow conditional branching, looping, and subprogram calls.
- Object level programming allows robot positions to be generated relative to items in the workplace, thus eliminating the need for the user to perform complex positional calculations.
- Dynamic collision checking facility automatically generates a detailed report of collisions between any two objects in the cell. This enables modifications to be made to the overall design before any equipment is installed, thus saving both time and money.
Examples of 3D Simulation in the Nuclear Industry
- WAGR Decommissioning Manipulator [5]:
The Wind scale Advanced Gas Cooled Reactor (WAGR) situated in Cumbria, England, was needed to be decommissioned by the United Kingdom Atomic Energy Authority (UKAEA). Methods of dismantling the pressure vessel had been developed, and the concept used a computer-controlled manipulator, equipped with thermal cutting tools, which is deployed from a rigid mast supported beneath a biological shield plate.
In order to design a manipulator, which could meet the specification it was first necessary to determine the kinematic configuration of the proposed machine. This was done using the GRASP package. The reactor vessel was simulated and various manipulator options explored. Particular points of interest were the ability to fold up the machine into the smallest package for withdrawal through the shield plate, and the requirement to operate inside a confined space yet maintaining normal tool attitude to the ‘hot box’ surface.
The results of the simulation helped in developing a traditional arm with shoulder, elbow and wrist movements, as the layout with its revolute joints, which, was well suited to working in, confined spaces.
- Process Vessel Inspection Manipulator [5]:
The Magnox fuel dissolver vessel used within the chemical separation plant at BNFL Sellafield was subjected to a set of periodic inspections. This vessel carried corrosive acid and at times it needed to be repaired due to leakage problems. Man access to the vessel was impossible due to high radiation level. The vessel contained a number of pipe connections and multiple branches around the hemispherical dome of the dissolver such that a manipulator with a dexterous reach was required. Extensive use of the GRASP package was made to optimize the configuration of the manipulator arm and build a solid model of the vessel and the pipe branches. Simulation of this arm required a routine to be written, as its movement is an epicyclic rotation.
A model manipulator was constructed and checked inside a mock-up of the dissolver, to confirm the validity of the GRASP simulation work. The manipulator was an integral part of a very complex remote handling system, and the simulation work provided the following insights:
- To design the arm configuration by traditional methods would have been very time consuming and expensive and GRASP has been very useful in reducing the design time.
- 3D graphical simulation is a very cost-effective way of designing and evaluating automated manufacturing systems. It can also help to ensure that new manufacturing systems operate as efficiently and productively as possible.
- A mockup of the final design in the hostile environment of the nuclear industry is necessary in order to save time and costs.
- The design stage of a new manipulator can be very long if all possible movements of the joints have to be drawn using 2D drafting techniques.
STUDY OF LOAD BALANCING FOR JUST-IN-TIME OPERATIONS
With the shift towards Just-In-Time manufacturing, many factory assembly lines are restricting the level of work in process inventories permitted between successive operations. This limit is usually accomplished by a combination of facility design and appropriate operational rules.
Regardless of the precise mechanism employed to accomplish inventory controls, the performance of production lines can be adversely affected. This can be due to "starvation" (a machine being idle because its input buffer is empty) or "blocking" (a machine being idle because its output buffer is full).
One of the ways to effectively tackle the above problem is the product available for immediate entry into the line should be loaded in an order, which avoids excessive idle time. Jobs of different product type have different processing requirements, setup times and routings. Computer simulation, therefore, is crucial for achieving the best order [9].
A detailed simulation model of feed-forward flow lines, written in the SIMAN language, was used to evaluate the effectiveness of the flow line consisting of seven work centers. Each center contained one or more parallel servers capable of performing the same operation. An external loading queue is used to place new jobs prior to being transferred to their initial work center.
Two simple approaches were used for comparison purposes [9]:
- Random by job: The set of jobs is randomly permuted.
- Random by code: All of the same type or code are forced to be adjacent in the sequence but the codes are randomly permuted (all jobs of a given code are not constrained, however, to travel together through the line in the model).
Several factors, which can characterize a production line, were considered in building the model. These included: overall utilization, processing time variability, setup intensiveness, line configurations, routing patterns and daily product delivery.
The results of the simulation model were plotted and a series of curves were plotted. The results were analyzed and following conclusions were made [9]:
- If setup times are not significant, grouping jobs by code can cause very poor performance due to substantial load imbalance caused by large numbers of jobs of the same code being adjacent in the sequence.
- Ignoring setup times as random-by-job sequence can be equally poor for setup intensive lines. (In this case setup efficiency outweighs the better load balance produced by randomly spreading out the jobs).
- A smaller percentage of actual setup time does not always result in better overall line performance.
- The simulation program developed adjusted to the wide range of setup time and utilization conditions likely to be prevalent in a large manufacturing company. It gave more weight to load balance when setup times are small and more weight to setup efficiency when they are larger. It therefore effectively balanced these two important concerns.
- The simulation model substantially improved the performance of a line relative to random sequencing- especially under conditions of high setup times or utilization. Combining it with appropriate rules for dispatching and lot sizing can further enhance its value.
SIMULATION IN VEHICLE MANUFACTURING
Flexibility to the market, production volume and assortment changes are becoming more and more important for industrial manufacturing systems. Under such conditions coordination of the production system becomes more complicated. There are many factors which affect the production process: absence of labor force, undermined manufacturing times, equipment failures, lack of work or blockage and carry-over effects of all these on other workstations (integration effects). In many cases simulation seems to be an appropriate tool for interactive design [10].
Simulation was applied to design and simulate the Toyota production systems factory. The goal of this article is to assess the benefits of simulation approach compared to traditional methods in vehicle manufacturing.
SIMFACTORY program was chosen as a simulation tool because of the following reasons [10]:
- Easy and fast to use
- Animation possibility and suitable for layout design
- Availability of customized reports
The machine manufacturing is divided in module making and assembly factories. A forwarder consists of three basic modules: frame parts, cabin and loader. A multifunction machine includes also a multifunction head, called aggregate.
All the three module-manufacturing units share one common plate shop. The plate shop includes an automatic plate store, stations for thermal NC-cutting and mechanical plate cutting, NC edging machines and radial drills and eccentric presses for auxiliary operations [10].
The manufacturing lines are fairly similar to each other: Robot welding is carried out first. Then the parts are machined with NC-machines, which form flexible manufacturing modules. Next the parts are painted and brought to the module assembly stations. From there the parts are further transported to the main assembly line.
The final assembly is organized as follows: The front and rear frames are assembled first in separate assembly lines. For example engines, driving shafts and hydraulics are mounted on these lines. Then the frames are joined together and the rest of the components, loader, cab, aggregate etc. are mounted. The cab frames are taken to the assembly factory where the equipments are mounted as a side assembly [10].
SIMULATION MODEL
In building the model the whole manufacturing is simulated from raw materials to a finished product. Production is separated to module manufacturing and to assembly. All the modules go through a similar manufacturing chain: first plate works, then welding and machining. From machining the parts go to painting and after that to module assembly.
Ready modules are taken to the assembly department. Cabin frames come to the assembly line and they are equipped there as a side assembly. From the main assembly line the forest machines go to testing and finishing [10].
It took about 8-9 hours for an experienced SIMFACTORY user to create a model of this size including preliminary planning. A simulation run of 500hrs with IBM PS/50 took 45-180 minutes depending on running parameters, especially on amount of inputs to the production system. If simulation model is run with animation, it takes approximately ten minutes more time, because updating graphics on screen loads the computer heavily [10].
Several curves were plotted and the following results were obtained:
- The curve of Work in Process (WIP) against capacity indicated that a slight increase of WIP raises the utilization rate strongly, but the rising becomes slow and finally the utilization rate begins to decline.
- The curve of Load against Costs (WIP and work) indicated that if the amount of jobs started is more than the planned capacity, the throughput doesn’t increase considerably, only the WIP level increases. But if the amount of work orders is decreased, the throughput decreases slower.
- WIP increases linearly with batch size.
- The throughput increases with batch size, because the share of setting time decreases. The curve gets flatter when the batch size is increased more. Increased amount of WIP and also integration effects between machines cause disturbance in the process.
- The due time against batch size curve indicated that the due time increases first but then the curve gets flatter because the share of setting time decreases and compensates partly those disturbances caused by increased WIP.
The simulation program was easy to use and the generated results were sufficient for comparative analysis. The simplifications made during the model-building phase did not affect the accuracy of the final results. The model functioned as intended and represented the actual factory layout.
Therefore, the simulation model was helpful to illustrate interesting dependencies and bottlenecks in the difficult cases where quantitative accuracy was difficult to reach [10].
STUDY AND REALISATION OF A MANUFACTURING SCHEDULER
Effective production scheduling has been an important area of concern for many years in manufacturing [1].
One of the major objectives of Production Planning and Control (PPC) is summarized as follows:
- Maximize concurrency of operations and minimize conflicts.
- Determine the sequence of operations leading to the completion of each job.
- Determine the timing of these events.
- Check material availability and update job priorities.
- Minimize production cost and lateness.
FACTOR is the most widely used software used in the manufacturing industries to achieve the PPC objectives. It is a discrete event simulation based finite scheduling tool, which generates achievable schedules using scheduling philosophy rules and a model of the physical production system.
The use of FACTOR has helped in improving the following project aspects [1]:
- Quality of detailed schedules.
- The flexibility and maintainability of the scheduling application with regard to the use of existing data.
- The ability to react to common changes in the production system environment.
- The ability of the application to be used by individuals for which, it is designed with a minimum amount of training.
- The regularity, timing and speed with which the application can be executed.
Scheduling organizes the shop for producing the parts and assemblies, and a data collection system is the source of information for scheduling and order release. Therefore, the key to shop floor management and control is a comprehensive system that links higher level planning and order release, detailed scheduling, and production process data collection [1].
The ability to simulate a series of actions or circumstances, to change them and observe the consequences and, therefore to increase confidence in certain actions is growing in the modern age of computers. Although large, expensive and very capable packages are available to simulate and analyze highly complex situations, the challenge facing the software developers is to build small, powerful and user-friendly packages. The simulation industry is coming of age. Soon those companies who do not update, implement simulation software would struggle to keep pace with the immense competition [1].
Back to Table of Contents